Predictive value of specific risk factors, symptoms and signs, in diagnosing obstructive sleep apnoea and its severity.

A positive diagnosis of obstructive sleep apnoea (OSA) is based on a combination of characteristic symptoms and polysomnographic findings. The present study evaluated the specificity and sensitivity of several risk factors, signs and symptoms in predicting an Apnoea Index in 86 patients referred to the sleep laboratory with suspected OSA. All 86 subjects completed a detailed questionnaire, were interviewed, underwent a brief physical examination, and then a whole-night polysomnographic study. Stepwise multiple regression analysis revealed that self reporting on apnoeas, neck circumference index (NCI), age, and a tendency to fall asleep unintentionally, were all significant positive predictors of apnoea index (AI), explaining 41.8% of the variability. The sensitivity of the model for predicting OSA (taking OSA as AI > 10) was 92.2%, specificity was 18.2% and the positive predictive value was 76.6%. Raising the cut-off AI values resulted in decreased sensitivity and increased specificity. Applying the predicting equation of AI to another group of 50 patients referred to the sleep laboratory with suspected OSA revealed similar results. However, running the equation on 105 offspring of OSA patients who did not complain of OSA-associated symptoms resulted in 32% sensitivity and 94% specificity in predicting OSA. It is concluded that questionnaires, interviews and physical examination, can only vaguely predict AI, and cannot replace polysomnographic recordings. However, the low rates of false negative in predicting AI > 10, and the low rates of false positive in predicting AI > 50, can be used for specific purposes.